Building Report: Goodwin Hall Engineering Lab Report

Building Report: Goodwin Hall
Student Name
Course
Institutional Affiliation
Executive Summary
The condition of Goodwin Hall during its lifespan needs to be assessed in order to preserve its structural integrity. Therefore, a polynomial curve fit model was validated and developed to determine the wind loading effects in Goodwin Hall. This encompasses applications for both present and future structural health monitoring (SHM). Since we found a visual correlation between the root mean square (RMS) wind acceleration and RMS wind velocity, we considered 52 minutes as the optimal window size. This was in comparison to window sizes of 4 and 345 minutes. The optimal model for both Mode 3 and Mode 1 was the third order of the polynomial curve fit model considering a window size of 52 minutes. This was arrived at by comparing the RMSE, Validation RMSE, and R-squared value. Further studies using the same evaluation procedure need to be conducted for higher order polynomials.
Introduction
Goodwin Hall is the College of Engineering’s flagship building in Virginia Tech. Within its halls are 150 offices for engineering departments, 40 research and instructional labs, eight classrooms, in addition to the Quillen Family Auditorium [1]. More than these, however, Goodwin Hall is an experimental building that is designed to measure the smallest vibrations within. The Hall is designed as a test bed that can track data related to its security and design, structural health monitoring, and occupancy for emergency response. There are roughly 240 accelerometers that are attached to 136 mounted sensors throughout the ceiling of the building and can detect information on the location of people within the structure, measure wind loads and normal structural settings, and to track the movement of the building that result from earthquakes [2].
A SHM system is built into the building, which was developed by Virginia Tech Smart Infrastructure Lab (VTSIL). The purpose of this SHM is to evaluate the structural integrity of the building based on modal responses and vibrational records. When the SHM detects vibrations that deviate from the range of normal variations, researchers in the Goodwin Hall will be alerted and they can make rapid changes to the building, thus avoiding severe damage [1]. However, changes to the vibrational measures can also be brought about by man-made and environmental factors. For instance, extreme temperature changes, strong winds, and extreme humidity can affect the building’s structural response. In this experiment, a polynomial curve fit model was derived from focused and analyzed wind loads which was able to estimate the influence of these factors on the vibrational response of Goodwin Hall. Through this method, false positive results can be avoided.
The present experiment is important because assessing the structural integrity of any building is paramount to assuring its safety. Vibrational changes that occur over the course of time can damage any building. Thus, this experiment is important for safety reasons, especially in response to natural and man-made disasters.
During a pre-defined time, wind data was collected from a building sensor. Wind data, comprised of wind speed and wind acceleration, was taken from a weather station found in the Lane Stadium. The data from the weather station was collected in 1-minute intervals with a frequency of 0.167. Data from the Goodwin Hall was collected very 0.125 seconds with a frequency of 10 Hz. The data taken from these two sources were divided into two sets in order to increase the validity of the predictive quality of the resulting polynomial curve model. Polynomial fit curve models with the property of increasing polynomials form 1 to 5 can be developed based on the data that was gathered. In the experiment, 70% of the data was used for training while 30% of the data was used for validation in order to justify the polynomial curve fit model. This validation process required statistical values such as root mean square error (RMSE), R-squared, validation root mean squared error (VRMSE). These were analyzed to determine the optimal values for the polynomial curve fit model.
Methods
Data Processing
Using the RMS approach, the MATLAB graphic user interface (GUI) was used to extract data on the vibrational energy during the pre-determined time. The RMS is a measure of the energy of a signal. This gives a scale that can be representative of the values sampled during the time window. The GUI is shown in Figure 1, whereas the time period in Figure 2 between K1 and K2, as well as the RMS value, can be calculated by the following equation:
yrms=1Nk1k2y2[k]
Here, N represents the total number of samples from the period between K1 and K2. Time windows and time intervals were used to arrive at the RMS values. The window speed measurement is precise by half a mile; thus, the data was rounded up by half a mile per hour. The time interval values were not utilized for this analysis since the collected data has back and forth oscillations around zero. This, correspondingly, gives a mean of zero, which is irrelevant for further analysis. In order to get the most precise and representative value, the RMS approach was used for wind information. This method also makes up for the precision that was not accounted for during measurement and it also has the effect of smoothing out the curvatures of the signal.
Figure 1. GUI with RMS Sample Window Size 1 min and Bandpass Filter Bounds 1 to 3.5 Hz.
Figure 2. RMS Window Size and Bandpass Filter Bounds
The GUI has an application for band-pass filters and can distinguish between the acceleration data to allow selected frequencies to pass. In Figure 3, there are three natural frequencies that correspond to the modes found in Goodwin Hall. Mode 1, which has a frequency of 1.88 Hz, corresponds the building bending to the north and south direction. Mode 2 represents the building bending in the east and west direction. Mode 3, which has a frequency of 2.3 Hz, represents the torsion of Goodwin Hall. Since a uniaxial north-facing accelerometer was used that was perpendicular to the east and west direction, Mode 2 cannot be observed in the collected data shown in Figure 4. The band-pass filter in the GUI was used as an application for the frequency boundaries for both Mode 1 and Mode 3.
Figure 3. Goodwin Hall ambient vibrations (Three natural frequencies)
Figure 4. Goodwin Hall power spectrum showing Modes 1 and 3
The RMS window sizes were randomly designated as 4, 345, and 52 minutes. As the window size increases, the number of samples likewise decreases. A short time interval, designated by a small window size, generates excessive data. This can cause noise from the frequently changing values. Figure 5 shows that the 4-minute window size generated 2355 samples, and this can cause the presence of too many outliers. Even though excessive sample sizes provide for a better polynomial curve fit model, at the same time, it causes an inaccurate fitting performance. There were only 27 samples in the 345-minute window size. Here, the small sample size makes it difficult to determine the relationships between the wind speed and the wind acceleration. Because of the extremes in sample sizes generated between the 4-minute window and the 345-minute window, 52 minutes emerged as the optimal window size because there were 183 samples and a relationship between the wind speed and wind acceleration can be seen. Statistics such as RMSE, VRMSE, and Re-squared will be compared for val...